How to run a gold DCA plan: a workable 500 USD a month

"Dollar-cost averaging" is easy to say and hard to do. The moment you go to place the order the questions get specific: where do I buy, how often, how much a month, do I stop when it falls, when do I sell. There are no standard answers, but every one of them has a cost you can calculate.

This article is arithmetic, not mindset. Using today's live quotes it breaks down a year of costs across three workable routes; it replays DCA vs lump sum over the real closing prices of the past 27 trading days; and it runs 5,000 simulated paths to answer a long-argued question — how much does weekly buying really differ from monthly?

Data captured 2026-10-20 20:24 (Beijing time) · Quotes: XAUAD live feed · FX: public rate API

While we are here, the conversion: 4,121.43 ÷ 31.1035 × 6.6937 = 886.96 per gram, i.e. 27,587 per ounce or 886,960 per kilogram. That number is the local-currency reference for everything below. Over the last 27 trading days gold fell from 4,392.33 to 4,120.00, a range of -6.2%, with a high of 4,402.73 and a low of 4,073.93, an average daily range of 1.42% and realised annualised volatility of about 19.6% — hold on to those three numbers, the simulation later uses them.

1. What DCA solves is not "buying cheap"

Most people think the point of DCA is lowering your average cost. What it really solves is a probability problem: a lump sum has exactly one entry price, while DCA spreads the entry across n prices. You did not become smarter; you moved the chance of "buying the exact top" from 100% to 1/n.

The price of that? If the market rises all the way, DCA's average cost is guaranteed to be higher than buying everything on day one. DCA is never a return-maximising tool; it is an error-minimising tool — it gives up the best outcome in a rising market to remove the worst outcome in a falling one. Understand that and you will not lose faith during a one-way rally, nor panic and stop buying in a selloff.

One more layer people miss: DCA only helps on volatile assets. Gold's annualised volatility sits in the 15%–20% band year after year (currently 19.6%) — far above deposits, below small caps and crypto, which makes it a "worth averaging into, but no need to watch daily" asset. If you are averaging into a money-market fund, frequency and cost are both irrelevant.

2. Three workable routes

For overseas Chinese there are really only three ways to do this. They differ not in whether you can buy, but in how much is deducted every time money goes in.

Note: fee ranges are typical industry levels; the published terms of your broker, dealer or platform govern.

The three routes are three completely different holding experiences: an ETF is "a number in an account", physical is "weight in a safe", spot is "floating P&L in a margin account". Pick the form you can hold for three years without touching, not the one with the lowest fee — the cheapest option you cannot hold is worth zero.

3. Costs in dollars: 500 USD a month, 6,000 USD a year

Assume 500 USD a month for twelve months, 6,000 USD invested. At today's 4,121.43 USD/oz that is about 1.46 oz, or 45.3 grams of gold. Here is what each route actually costs:

Put the three side by side and the conclusion is blunt: an ETF costs about 10 USD a year, spot about 91, physical from 240. The gap is not about platforms being predatory, it is a different cost structure — an ETF folds custody into a 0.25% annual fee, spot folds the cost of money into a daily financing charge, physical folds refining, minting and dealer margin into a one-off premium.

Overnight financing is the hidden tax on DCA

A long spot gold position pays overnight interest, typically 2%–4% a year of notional (it tracks USD rates). At 1:1, with 6,000 USD of capital and an average holding of 3,000, that is about 90 USD a year; at 5× leverage notional becomes 15,000 and the financing is about 450 USD a year, or 7.5% of capital. Which is why DCA and leverage are mathematically contradictory: DCA bets on holding for years, and leverage charges rent precisely for holding.

4. Frequency: weekly, fortnightly, monthly — does it matter?

This is the least meaningful and most heated argument in DCA circles. The answer: next to cost, frequency barely matters.

I ran 5,000 random paths at the current realised volatility (19.6% annualised, consistent with the past 27 trading days) and compared the final cost of "buy weekly" against "buy monthly" on the same price path:

• The median cost difference is -0.04% (weekly marginally cheaper, negligible)
• In 83% of cases the difference falls within ±1%
• 10th percentile -0.98%, 90th percentile +0.86% — even the tails stay around 1%

Yet the number of trades rises from 12 to 52, and cost rises by a hard 4.3 times: ETF spreads, physical shipping and minimum fees, spot spreads are all charged per transaction. So the correct order is: compare cost first, argue frequency second.

If you are buying physical bars, quarterly actually wins: a 1 oz coin carries a lower premium than a 1/10 oz coin, and rounding up to whole ounces saves 1–3 percentage points of minting cost. That is the only reason to optimise frequency with physical gold.

5. Real replay: the last 27 days, DCA vs lump sum

Replaying the actual closes from 20 September to 20 October. Same 6,000 USD, three entry methods:

The result is interesting: DCA beat "lump sum at the high" by 4.5% (0.062 oz more gold), and lost to "lump sum at the low" by 2.0%. Over those 27 days gold fell 6.2% — had you known on 20 September, the optimal move was obviously to wait until 20 October. Nobody knows.

6. Setting the amount: fix the percentage, then derive the monthly debit

The order cannot be reversed. Decide what share of your financial assets gold should be, then divide the annual contribution by twelve. Doing it the other way (picking "500 USD a month" first) produces two classic errors: too big a share to hold, or too small to matter.

Two cash-flow rules matter more than the percentage: never commit more than 30% of your monthly surplus, and keep six months of living expenses in cash. The reason is practical — if DCA eats your emergency fund, one car repair or one job loss forces you to sell at the low, and once the discipline breaks it rarely comes back.

If you already have idle cash (say 50,000 USD) to allocate to gold, you do not have to enter all at once either. Split it across 6–12 months: in effect you trade "timing risk" for "time cost". Gold pays no interest, so the opportunity cost of staging entries is lower than for equities — one reason it suits slow accumulation.

7. Taking profit and rebalancing: two mechanical rules

Gold has no cash flow, no P/E and no dividend, so every "valuation anchor" is weak. Which is why the sell trigger in a gold DCA plan should not be "I think it is high", but deviation from target weight.

Rule one: rebalance at a ±5 percentage point band

Target 10%: when the actual weight rises above 15% or falls below 5%, rebalance once. Example: a 100,000 USD portfolio with a 10,000 USD gold target; if it grows to 15,000 (15%), sell 5,000 to get back to 10%; if it falls to 5,000, add 5,000. This rule takes profit and adds on weakness in one move, with no need to forecast.

Rule two: check annually, act at most twice a year

The more often you rebalance, the more a long-term allocation turns into short-term trading. Once a year, acting only when the deviation exceeds five percentage points, is the balance point. Frequent rebalancing bleeds cost on volatile assets — especially physical gold, where every sale means taking a dealer's buy-back discount.

8. For overseas Chinese: accounts, remittance, tax

Accounts: pick one you can hold for a decade, not the one with the best promotions

ETFs go through a locally licensed broker, physical through a local dealer that offers buy-back, spot through a platform regulated by the FCA, ASIC, CySEC or similar. Three bottom lines: client funds segregated, deposits and withdrawals in your own name, licence verifiable. DCA is a ten-year project; whether a platform will still exist in ten years matters far more than the first-month bonus. For how to check, see choosing a gold trading platform: 10 checks.

Remittance: quota, purpose, structuring

Moving money out of the mainland to invest must go through compliant personal channels: the annual foreign-exchange facilitation quota is the equivalent of 50,000 USD, the purpose must be declared truthfully, and it may not be used for capital-account items such as overseas property or securities investment. Regulators define structuring explicitly — five or more different individuals buying FX on the same day, the next day, or over consecutive days and remitting to the same recipient; once flagged, the quota for that year and the next may be withdrawn. Overseas banks will also ask for source-of-funds evidence on large deposits. Details in remitting money for gold: compliant routes and red lines.

Tax: follow the rules where you live, not the rumours

• United States: physical gold and gold ETFs are treated as collectibles, with long-term capital gains taxed at up to 28% after a year, above the 20% standard long-term rate
• Canada: 50% of capital gains is included in taxable income at your marginal rate
• Australia: assets held over 12 months qualify for a 50% CGT discount; investment gold and silver of 99.5%+ purity is GST-free
• United Kingdom: reported under capital gains tax, with rates tracking your income band; check the annual allowance yourself
• Hong Kong / Singapore: no capital gains tax, though high-frequency activity may be treated as trading income and taxed accordingly

And do not forget the FX layer: if your income and spending are both in USD, judging P&L in CNY terms uses the wrong coordinate system. Only when you will ultimately convert back to spend does the local-currency gold price become your real return.

9. The five most common mistakes

1. Treating DCA as capital protection. DCA changes the distribution of entry prices, not the direction of the asset. If gold falls for two years, DCA loses too.
2. Stopping when it falls. The most fatal one. Stopping throws away the single advantage — accumulating units at low prices — and leaves you with "buy high, watch low".
3. DCA-ing with leveraged spot. Financing is charged on notional: at 5× that is 7.5% a year, more expensive than any "low-cost platform" marketing claims.
4. Constantly changing amount or instrument. ETF this month, physical next month, a new platform the month after: three layers of cost and not one ounce of discipline.
5. Watching only the USD price. Ignoring local FX and local premiums leaves your calculated return several points away from what actually lands in your account.

10. FAQ

Q. How much should I put into gold each month?
A. Fix the percentage first and work backwards: gold at 5%–15% of financial assets, so 100,000 USD targeting 10% means 10,000 a year, about 833 a month. Then respect the two cash-flow rules — no more than 30% of monthly surplus, and six months of living expenses held in cash.

Q. ETF or physical bars for a gold DCA plan?
A. At 500 USD a month, 6,000 a year, an ETF costs about 10 USD for the year and physical about 240–360 — a twenty-to-thirtyfold difference. Choose the ETF for long-term preservation; pay the premium only if you want the physical security of metal you can put away.

Q. Should I keep buying when the price falls?
A. Yes. DCA's return comes from accumulating more units at low prices; stopping in a decline turns it into momentum chasing. There are only two legitimate reasons to stop: your cash flow broke, or you hit your target weight.

Q. How long before a gold DCA plan shows results?
A. A full gold cycle is typically three to five years, and one year of price action says nothing. A sensible window is three years plus, and judge it by whether the weight has drifted from target, not by the floating profit in the account.

Q. What should overseas Chinese watch for when DCA-ing into gold?
A. Be able to document where the money came from, remit through compliant personal channels (annual quota of 50,000 USD equivalent, honest declaration, no splitting), and report tax where you live. Only with all three can the plan actually run long term.

Related tools